Multi-worker occasionally does not balance on a small queue

[nerdalert]

Observed two longish running jobs (generate-local) from two different PRs not get balanced over the workers. Check the redis queue for details while both jobs are outstanding to tshoot. Balancing is occurring but this should be investigated.